Here's an editorial in today's Independent from one of the authors of the Lancet report. He offers a few observations that, if true, suggest that official counts are not very reliable:

[url]http://comment.independent.co.uk/commentators/article2268067.ece[/url]

Does that article clarify? Or does it illustrate the biases of one of the study's authors?

[quote=masser;98538]Does that article clarify? Or does it illustrate the biases of one of the study's authors?[/quote]It illustrates the bias of one of the authors [I]toward getting an accurate count[/I], as contrasted with the biases of Iraqi, U.S. and U.K. governments toward understating the toll of war, and clarifies some reasons for the undercounts preferred by Iraqi, U.S., and U.K. governments.

Some quotes from the article:

[quote=Les Roberts]The government in Iraq claimed last month that since the 2003 invasion between 40,000 and 50,000 violent deaths have occurred. Few have pointed out the absurdity of this statement.

There are three ways we know it is a gross underestimate. First, if it were true, including suicides, South Africa, Colombia, Estonia, Kazakhstan, Latvia, Lithuania and Russia have experienced higher violent death rates than Iraq over the past four years. If true, many North and South American cities and Sub-Saharan Africa have had a similar murder rate to that claimed in Iraq. For those of us who have been in Iraq, the suggestion that New Orleans is more violent seems simply ridiculous.

Secondly, there have to be at least 120,000 and probably 140,000 deaths per year from natural causes in a country with the population of Iraq. The numerous stories we hear about overflowing morgues, the need for new cemeteries and new body collection brigades are not consistent with a 10 per cent rise in death rate above the baseline.

And finally, there was a study, peer-reviewed and published in The Lancet, Europe's most prestigious medical journal, which put the death toll at 650,000 as of last July. The study, which I co-authored, was done by the standard cluster approach used by the UN to estimate mortality in dozens of countries each year. While the findings are imprecise, the lower range of possibilities suggested that the Iraq government was at least downplaying the number of dead by a factor of 10.
[/quote]Will one of you who contend that the Lancet estimate is too high please explain to us why South Africa and Colombia have had, supposedly, higher rates of violent death than Iraq for the past four years?

[quote]Repeated evaluations of deaths identified from sources independent of the press and the Ministry of Health show the IBC listing to be less than 10 per cent complete, but because it matches the reports of the governments involved, it is easily referenced.

Several other estimates have placed the death toll far higher than the Iraqi government estimates, but those have received less press attention. When in 2005, a UN survey reported that 90 per cent of violent attacks in Scotland were not recorded by the police, no one, not even the police, disputed this finding. Representative surveys are the next best thing to a census for counting deaths, and nowhere but Iraq have partial tallies from morgues and hospitals been given such credence when representative survey results are available.[/quote]

(Anyone want to bet that the anti-Lancet faction here will try to make something out of the difference between non-war Scottish violent attacks and war-zone Iraqi violent deaths, to draw attention away from the shared phenomenon of underreporting?)

[QUOTE=masser;98538]Does that article clarify? Or does it illustrate the biases of one of the study's authors?[/QUOTE]

Since it is a discussion in a newspaper, rather than technical notes in a scientific journal, it doesn't really clarify things much. From an article I read in the journal Science, however, including the follow-up letters and responses that described the methods for selecting which residences were included in the survey (how streets were selected, and how houses on those streets were selected, etc., etc.), the survey itself did not sound biased. I'm not a statistician, and have never studied the proper methodology of these kinds of surveys, but I don't see how the Lancet survey would have overestimated deaths by even twice the actual number, much less the 10X overestimation some people claim.

Norm

Cheesehead,

Don't forget the punchline:

[CODE]
Saddam Hussein's surveillance network, which only captured one third of all deaths before
the invasion, has certainly deteriorated even further. During last July, there were numerous
televised clashes in Anbar, yet the system recorded exactly zero violent deaths from the
province. The last Minister of Health to honestly assess the surveillance network, Dr Ala'din
Alwan, admitted that it was not reporting from most of the country by August 2004. He was
sacked months later after, among other things, reports appeared based on the limited
government data suggesting that most violent deaths were associated with coalition forces.


The consequences of downplaying the number of deaths in Iraq are profound for both the UK
and the US. How can the Americans have a surge of troops to secure the population and
promise success when the coalition cannot measure the level of security to within a factor of
10? How can the US and Britain pretend they understand the level of resentment in Iraq if
they are not sure if, on average, one in 80 families have lost a household member, or one in
seven, as our study suggests?


If these two countries have triggered an episode more deadly than the Rwandan genocide,
and have actively worked to mask this fact, how will they credibly be able to criticise Sudan
or Zimbabwe or the next government that kills thousands of its own people?


For longer than the US has been a nation, Britain has pushed us at our worst of moments to
do the right thing. That time has come again with regard to Iraq. It is wrong to be the junior
partner in an endeavour rigged to deny the next death induced, and to have spokespeople
effectively respond to that death with disinterest and denial.

Our nations' leaders are collectively expressing belligerence at a time when the populace
knows they should be expressing contrition. If that cannot be corrected, Britain should end
its role in this deteriorating misadventure. It is unlikely that any historians will record the
occupation of Iraq in a favourable light. Britain followed the Americans into this débâcle.
Wouldn't it be better to let history record that Britain led them out?
[/CODE]

My only point is that a statistician should do his best to present the numbers; not spin the numbers. He accuses others of "spinning science" and yet he appears to be doing the same. Also, have you ever head of the debating rule: "The first one to mention Hitler/Nazis loses the debate?" I think this guy is making a similar mistake by mentioning genocides.

[quote=Prime95;98375]When the surveys are counting different things violent deaths vs. excess deaths. I would accept 2-5x.[/quote](A) Does that mean you think that excess deaths by disease or illness outnumber excess deaths by violence by a factor of 1-4x?

(B) Are there possible significant causes of excess (i.e., compared to before the war) nonviolent deaths since start of the war that are not war-related, but just happen to be coincident in time? (If you argue that something like poorer health care or sanitation is responsible, then you also have to explain why they're so dramatically poorer [I]without having been made so by the effects of war[/I] despite the coincidence in timing.)

[quote=masser;98542]Cheesehead,

Don't forget the punchline:[/quote]I find it curious that you quote so much that precedes the author's opinion punchline, but without disputing it and without claiming that any of the numbers or nonnumerical facts presented are incorrect. Does that mean you do not contest any of the numbers or nonnumerical facts presented by the author?

[quote][code]Saddam Hussein's surveillance network, which only captured one third of all deaths before
the invasion, has certainly deteriorated even further.[/code][/quote]... which supports the author's contention that government figures are too low.

[quote][code]During last July, there were numerous
televised clashes in Anbar, yet the system recorded exactly zero violent deaths from the
province.[/code][/quote]... which supports the author's contention that government figures are too low.

[quote][code]The last Minister of Health to honestly assess the surveillance network, Dr Ala'din
Alwan, admitted that it was not reporting from most of the country by August 2004. He was
sacked months later after, among other things, reports appeared based on the limited
government data suggesting that most violent deaths were associated with coalition forces.[/code][/quote]... which supports the author's contention that government figures are too low.

[quote][code]The consequences of downplaying the number of deaths in Iraq are profound for both the UK
and the US. How can the Americans have a surge of troops to secure the population and
promise success when the coalition cannot measure the level of security to within a factor of
10? How can the US and Britain pretend they understand the level of resentment in Iraq if
they are not sure if, on average, one in 80 families have lost a household member, or one in
seven, as our study suggests?[/code][/quote]... which are unbiased questions about the connection between undercounts and policy.

[quote][code]If these two countries have triggered an episode more deadly than the Rwandan genocide,[/code][/quote]The word "genocide" is mentioned only twice in the article (in this quoted sentence and in the article's subtitle). In both cases, it is part of the phrase "the Rwandan genocide".

Would you prefer the author to have written "Rwandan incident" or "Rwandan troubles"? It wasn't really a war.

The author doesn't accuse or hint that either the US or UK committed genocide. He's simply pointing out that the death toll of the war initiated by the US and UK was greater than that in the Rwandan case/incident/bad-period.

[quote][code]and have actively worked to mask this fact, how will they credibly be able to criticise Sudan
or Zimbabwe or the next government that kills thousands of its own people?[/code][/quote]... and here, where the author could justifiably have used "genocide" again to describe events in another country (many news reports about Sudan that I've heard used the word "genocide"), he refrains.

[quote][code]For longer than the US has been a nation, Britain < snip >[/code][/quote]Finally, here the author expresses opinion. So? It's a newsmedia article, in the section titled "Commentators". Earlier, he expressed facts and comparisons. Now he goes into his opinions. Nothing wrong with that.

Since you didn't comment on any of the pre-opinion section, may we conclude that you don't dispute any of it?

[quote]My only point is that a statistician should do his best to present the numbers;[/quote]... which he did in his scientific report in the Lancet!

Here we have an article that doesn't present itself as an opinionless purely scientific report; it's in the "Commentators" section! The author is plainly pointing out connections between the data and his opinions.

Statisticians are allowed to have and express personal opinions. They shouldn't do so in scientific papers, but this is a commentary!

[quote]not spin the numbers. He accuses other of "spinning science" and yet he appears to be doing the same.[/quote]No. "Spin" tries to deceive. Where has the author deceived in this article? Nowhere. He doesn't present his opinion as fact; he presents connections between his opinions and facts.

Is there any number cited by the author in this article that is misrepresented as being anything other than what it actually is? Not that I can see, but perhaps my seeing is flawed. Can you point to any number in the article that is misrepresented as other than it really is?

[quote]Also, have you ever head of the debating rule: "The first one to mention Hitler/Nazis loses the debate?"[/quote]Yes, but this author mentions neither Hitler, Nazis, nor Holocaust.

[quote]I think this guy is making a similar mistake by mentioning genocides.[/quote]Actually, he mentioned only one, and that was one that I've heard, labelled with the same "Rwandan genocide" phrase, probably hundreds of times in news stories and conversations of the past decade. Again, how would you have the author refer to the Rwandan ... uh ... situation ... without using "genocide"?

- - - - - - - -

I can understand that supporters of Bush's military policy in Iraq may not like, be comfortable with, or readily accept some numbers and nonnumerical facts from the Lancet report, and may not ever accept or agree with personal opinions expressed (outside the report) by that report's authors. But can any of you point to actual deception or distortion of numbers or nonnumerical facts?

When, in a response to Prime95 about 105 minutes ago, I wrote:
[quote=cheesehead;98543](A) Does that mean you think that excess deaths by disease or illness[/quote]here I should have included "or by accident" before continuing
[quote]outnumber excess deaths by violence by a factor of 1-4x?[/quote]

Sorry, Cheesehead; I guess I wasn't being pedantic enough.

My point here is that after reading the article I question whether the author's statistics inform his opinion or if his opinion informs his statistics.

What I was trying to say about the author's genocide (singular; thanks for jumping all over a typo, btw) comment, was that by comparing death totals from the war in Iraq to the Rwandan genocide (singular, again), he's doing something similar to mentioning Hitler in a debate. I never said he mentioned Hitler...

And just to rile you up some more, a little skit:

Neocon: Because of 9/11, something something something, invade Iraq!
Cheesehead: Bad Neocon! You're linking 9/11 to Iraq! There is no connection!

Lancet Author: Death toll in Iraq, something something something, genocide in Rwanda!
masser: umm....

Lastly, a kind request: how does one put a quote within a quote on this forum? I'm feeling dumb...

masser

Quote:
Also, have you ever head of the debating rule: "The first one to mention Hitler/Nazis loses the debate?"

Yes, but this author mentions neither Hitler, Nazis, nor Holocaust.

Quote:
I think this guy is making a similar mistake by mentioning genocides.

Actually, he mentioned only one, and that was one that I've heard, labelled with the same "Rwandan genocide" phrase, probably hundreds of times in news stories and conversations of the past decade. Again, how would you have the author refer to the Rwandan ... uh ... situation ... without using "genocide"?

[QUOTE=cheesehead;98543](A) Does that mean you think that excess deaths by disease or illness outnumber excess deaths by violence by a factor of 1-4x?[/QUOTE]

If I understand your question, no.

Where the UN report only counts violent deaths and the Lancet report counts excess deaths from all causes including violence I would not find a Lancet number of 2-5x surprising.

[quote=masser;98563]Sorry, Cheesehead; I guess I wasn't being pedantic enough.

My point here is that after reading the article I question whether the author's statistics inform his opinion or if his opinion informs his statistics.[/quote]Okay.

Which number(s) or nonnumerical fact(s) in the article or in the Lancet report do you contest?

[quote](singular; thanks for jumping all over a typo, btw)[/quote]Had I known it was a typo, I'd have treated it (merely) as such. As it was, I didn't accuse you of deliberately multiplying the genocides; I just corrected a factual mistake because there was a possibility ([I]now[/I] known not to exist) that the plural number might play some role in some other mistake.

[quote]by comparing death totals from the war in Iraq to the Rwandan genocide (singular, again), he's doing something similar to mentioning Hitler in a debate.[/quote]I disagree that the similarity had significance.

Again, [I]how do you propose that the author refer to the Rwandan genocide without using the word "genocide"?[/I] If you can't come up with any reasonable answer to that question (and so far you haven't), you have no business accusing the author of committing the Hitler/Nazi debate-loser.

[quote]I never said he mentioned Hitler...[/quote]... nor did I say you did. You're the one who introduced "Hitler" and "Nazi" into the conversation, and I don't think you had sufficient justification to do so.

[quote]And just to rile you up some more, a little skit:[/quote]So, instead of straightforwardly answering my inquiries in my earlier posting about

(A) what fact(s) you contested or

(B) whether you dispute any of what I called the "pre-opinion section" or

(C) what word you would have the author use instead of what you imply is an unnecessary "genocide",

you present ... a little skit.

I'll be more impressed by straight answers and comments that actually pertain to what the Lancet report author actually wrote or what I asked you than I will be by any metaphor.

[quote]Cheesehead: Bad Neocon! You're linking 9/11 to Iraq! There is no connection![/quote]Do you mean that no neocon has ever linked 9/11 to Iraq?

If that's not why you attribute that dialog to me in your skit, please say straightforwardly, without indirect allusion or metaphor, just why you do attribute those words to me in your skit.

[quote]Lancet Author: Death toll in Iraq, something something something, genocide in Rwanda!
masser: umm....[/quote]Umm ... what? Spit it out, please.

- - -

Someone else please answer masser's question about putting a quote within a quote.